Expose ENCAP mode to mlx5_ib
Subsystems (5)
cover Expose ENCAP mode to mlx5_ib
From: Leon Romanovsky <[email protected]>
Hi,
This is short series from Maor to expose and use enacap mode inside mlx5_ib.
Thanks
Maor Gottlieb (3):
net/mlx5: Expose eswitch encap mode RDMA/mlx5: Consider eswitch encap mode RDMA/mlx5: Enable decap and packet reformat on FDB
drivers/infiniband/hw/mlx5/main.c | 25 ++++++++++++++----- .../net/ethernet/mellanox/mlx5/core/eswitch.c | 10 ++++++++ include/linux/mlx5/eswitch.h | 10 ++++++++ 3 files changed, 39 insertions(+), 6 deletions(-)
PATCH 1/3 net/mlx5: Expose eswitch encap mode
From: Maor Gottlieb <[email protected]>
Add API to get the current Eswitch encap mode. It will be used in downstream patches to check if flow table can be created with encap support or not.
--- drivers/net/ethernet/mellanox/mlx5/core/eswitch.c | 10 ++++++++++ include/linux/mlx5/eswitch.h | 10 ++++++++++ 2 files changed, 20 insertions(+)
drivers/net/ethernet/mellanox/mlx5/core/eswitch.c +10 -0
@@ -2452,6 +2452,16 @@ u8 mlx5_eswitch_mode(struct mlx5_eswitch *esw) }EXPORT_SYMBOL_GPL (mlx5_eswitch_mode); +u16 mlx5_eswitch_get_encap_mode (struct mlx5_core_dev *dev) +{ +struct mlx5_eswitch *esw; + + esw = dev->priv.eswitch; +return ESW_ALLOWED (esw) ? esw->offloads.encap : +DEVLINK_ESWITCH_ENCAP_MODE_NONE ; +} +EXPORT_SYMBOL (mlx5_eswitch_get_encap_mode); +bool mlx5_esw_lag_prereq (struct mlx5_core_dev *dev0,struct mlx5_core_dev *dev1) { if ((dev0->priv.eswitch->mode == SRIOV_NONE &&
include/linux/mlx5/eswitch.h +10 -0
@@ -7,6 +7,7 @@#define _MLX5_ESWITCH_#include <linux/mlx5/driver.h> +#include <net/devlink.h> #define MLX5_ESWITCH_MANAGER (mdev) MLX5_CAP_GEN(mdev, eswitch_manager)
@@ -60,4 +61,13 @@ u8 mlx5_eswitch_mode(struct mlx5_eswitch *esw);struct mlx5_flow_handle *mlx5_eswitch_add_send_to_vport_rule (struct mlx5_eswitch *esw,int vport,u32 sqn); + +#ifdef CONFIG_MLX5_ESWITCH +u16 mlx5_eswitch_get_encap_mode (struct mlx5_core_dev *dev); +#else /* CONFIG_MLX5_ESWITCH */ +static inline u16 mlx5_eswitch_get_encap_mode (struct mlx5_core_dev *dev) +{ +return DEVLINK_ESWITCH_ENCAP_MODE_NONE ; +} +#endif /* CONFIG_MLX5_ESWITCH */ #endif
Signature · 1 line
2.20.1
Re: [PATCH mlx5-next 1/3] net/mlx5: Expose eswitch encap mode
Hi,
Quoted 8 lines
From: Maor Gottlieb <[email protected]>
Add API to get the current Eswitch encap mode. It will be used in downstream patches to check if flow table can be created with encap support or not.
Signed-off-by Maor Gottlieb <[email protected]>Signed-off-by Leon Romanovsky <[email protected]>
Kind regards,
Petr
Re: [PATCH mlx5-next 1/3] net/mlx5: Expose eswitch encap mode
Quoted 11 lines · Petr Vorel
Hi,
Quoted 8 lines
From: Maor Gottlieb <[email protected]>
Add API to get the current Eswitch encap mode. It will be used in downstream patches to check if flow table can be created with encap support or not.
Signed-off-by Maor Gottlieb <[email protected]>Signed-off-by Leon Romanovsky <[email protected]>Reviewed-by Petr Vorel <[email protected]>
Thanks Petr for review, I'll add your tags and resend.
Quoted 3 lines
Kind regards,
Petr
RE: [PATCH mlx5-next 1/3] net/mlx5: Expose eswitch encap mode
In reply to Leon RomanovskyQuoted 33 lines
-----Original Message-----
From: [email protected] <linux-rdma- [email protected]> On Behalf Of Leon Romanovsky
Sent: Thursday, June 6, 2019 4:36 PM
To: Doug Ledford <[email protected]>; Jason Gunthorpe <[email protected]>
Cc: Leon Romanovsky <[email protected]>; RDMA mailing list <linux- [email protected]>; Maor Gottlieb <[email protected]>; Mark Bloch <[email protected]>; Saeed Mahameed <[email protected]>; linux-netdev <[email protected]>
Subject: [PATCH mlx5-next 1/3] net/mlx5: Expose eswitch encap modeFrom: Maor Gottlieb <[email protected]>
Add API to get the current Eswitch encap mode.
It will be used in downstream patches to check if flow table can be created with encap support or not.Signed-off-by Maor Gottlieb <[email protected]>Signed-off-by Leon Romanovsky <[email protected]>--- drivers/net/ethernet/mellanox/mlx5/core/eswitch.c | 10 ++++++++++ include/linux/mlx5/eswitch.h | 10 ++++++++++ 2 files changed, 20 insertions(+)drivers/net/ethernet/mellanox/mlx5/core/eswitch.c +0
-0
drivers/net/ethernet/mellanox/mlx5/core/eswitch.c u8 mlx5_eswitch_mode(struct mlx5_eswitch · line 2454
Quoted 3 lines
drivers/net/ethernet/mellanox/mlx5/core/eswitch.c +1
-0@@ -2452,6 +2452,16 @@ u8 mlx5_eswitch_mode(struct mlx5_eswitch *esw) } EXPORT_SYMBOL_GPL(mlx5_eswitch_mode); +u16 mlx5_eswitch_get_encap_mode (struct mlx5_core_dev *dev) {
Encap mode as well defined devlink definition.
So instead of u16, it should return enum devlink_eswitch_encap_mode.
Since this is only reading the mode, it is better to define struct mlx5_core_dev* as const struct mlx5_core_dev *.
Quoted 38 lines
drivers/net/ethernet/mellanox/mlx5/core/eswitch.c +8
-0+struct mlx5_eswitch *esw; + + esw = dev->priv.eswitch; +return ESW_ALLOWED (esw) ? esw->offloads.encap : +DEVLINK_ESWITCH_ENCAP_MODE_NONE ; +} +EXPORT_SYMBOL (mlx5_eswitch_get_encap_mode); + bool mlx5_esw_lag_prereq(struct mlx5_core_dev *dev0, struct mlx5_core_dev *dev1) { if ((dev0->priv.eswitch->mode == SRIOV_NONE && diff --git
a/include/linux/mlx5/eswitch.hb/include/linux/mlx5/eswitch.hindex0ca77dd1429c..7be43c0fcdc5100644include/linux/mlx5/eswitch.h +1
-0@@ -7,6 +7,7 @@#define _MLX5_ESWITCH_#include <linux/mlx5/driver.h> +#include <net/devlink.h> #define MLX5_ESWITCH_MANAGER (mdev) MLX5_CAP_GEN(mdev, eswitch_manager)@@ -60,4 +61,13 @@ u8 mlx5_eswitch_mode(struct mlx5_eswitch *esw);struct
mlx5_flow_handle* mlx5_eswitch_add_send_to_vport_rule(structmlx5_eswitch*esw,include/linux/mlx5/eswitch.h +7
-1int vport, u32 sqn); + +#ifdef CONFIG_MLX5_ESWITCH +u16 mlx5_eswitch_get_encap_mode (struct mlx5_core_dev *dev);#else /* +CONFIG_MLX5_ESWITCH */ static inline u16 +mlx5_eswitch_get_encap_mode (struct mlx5_core_dev *dev) { +return DEVLINK_ESWITCH_ENCAP_MODE_NONE ; }#endif /* +CONFIG_MLX5_ESWITCH */ #endif --2.20.1
Re: [PATCH mlx5-next 1/3] net/mlx5: Expose eswitch encap mode
Quoted 35 lines · Parav Pandit
Quoted 35 lines
-----Original Message-----
From: [email protected] <linux-rdma- [email protected]> On Behalf Of Leon Romanovsky
Sent: Thursday, June 6, 2019 4:36 PM
To: Doug Ledford <[email protected]>; Jason Gunthorpe <[email protected]>
Cc: Leon Romanovsky <[email protected]>; RDMA mailing list <linux- [email protected]>; Maor Gottlieb <[email protected]>; Mark Bloch <[email protected]>; Saeed Mahameed <[email protected]>; linux-netdev <[email protected]>
Subject: [PATCH mlx5-next 1/3] net/mlx5: Expose eswitch encap modeFrom: Maor Gottlieb <[email protected]>
Add API to get the current Eswitch encap mode.
It will be used in downstream patches to check if flow table can be created with encap support or not.Signed-off-by Maor Gottlieb <[email protected]>Signed-off-by Leon Romanovsky <[email protected]>--- drivers/net/ethernet/mellanox/mlx5/core/eswitch.c | 10 ++++++++++ include/linux/mlx5/eswitch.h | 10 ++++++++++ 2 files changed, 20 insertions(+)drivers/net/ethernet/mellanox/mlx5/core/eswitch.c +0
-0@@ -2452,6 +2452,16 @@ u8 mlx5_eswitch_mode(struct mlx5_eswitch *esw) } EXPORT_SYMBOL_GPL(mlx5_eswitch_mode);
Quoted 6 lines
drivers/net/ethernet/mellanox/mlx5/core/eswitch.c u8 mlx5_eswitch_mode(struct mlx5_eswitch · line 2454
Quoted 1 line
drivers/net/ethernet/mellanox/mlx5/core/eswitch.c +1
-0+u16 mlx5_eswitch_get_encap_mode (struct mlx5_core_dev *dev) {Encap mode as well defined devlink definition.
So instead of u16, it should return enumdevlink_eswitch_encap_mode.Since this is only reading the mode, it is better to define struct mlx5_core_dev* as const struct
mlx5_core_dev*.
Thanks Parav, I'll change and resend, anyway second patch uses wrong types too.
PATCH 2/3 RDMA/mlx5: Consider eswitch encap mode
In reply to Leon RomanovskyFrom: Maor Gottlieb <[email protected]>
When flow steering is created, then the encap support should consider the eswitch encap mode. If the eswitch flow table (FDB) supports encap then it shouldn't be supported on NIC RX flow tables.
--- drivers/infiniband/hw/mlx5/main.c | 20 ++++++++++++++------ 1 file changed, 14 insertions(+), 6 deletions(-)
drivers/infiniband/hw/mlx5/main.c +14 -6
@@ -52,6 +52,7 @@#include <linux/mlx5/port.h> #include <linux/mlx5/vport.h> #include <linux/mlx5/fs.h> +#include <linux/mlx5/eswitch.h> #include <linux/list.h> #include <rdma/ib_smi.h> #include <rdma/ib_umem.h>
@@ -3247,9 +3248,12 @@ static struct mlx5_ib_flow_prio *get_flow_table(struct mlx5_ib_dev *dev,int num_groups;u32 flags =0 ;int priority; +u8 esw_encap; max_table_size =BIT (MLX5_CAP_FLOWTABLE_NIC_RX (dev->mdev, log_max_ft_size)); + esw_encap =mlx5_eswitch_get_encap_mode (dev->mdev) != +DEVLINK_ESWITCH_ENCAP_MODE_NONE ;if (flow_attr->type ==IB_FLOW_ATTR_NORMAL ) {enum mlx5_flow_namespace_type fn_type;
@@ -3262,10 +3266,10 @@ static struct mlx5_ib_flow_prio *get_flow_table(struct mlx5_ib_dev *dev,if (ft_type ==MLX5_IB_FT_RX ) { fn_type =MLX5_FLOW_NAMESPACE_BYPASS ; prio = &dev->flow_db->prios[priority]; -if (!dev->is_rep && +if (!dev->is_rep && !esw_encap &&MLX5_CAP_FLOWTABLE_NIC_RX (dev->mdev, decap)) flags |=MLX5_FLOW_TABLE_TUNNEL_EN_DECAP ; -if (!dev->is_rep && +if (!dev->is_rep && !esw_encap &&MLX5_CAP_FLOWTABLE_NIC_RX (dev->mdev, reformat_l3_tunnel_to_l2)) flags |=MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT ;
@@ -3275,7 +3279,7 @@ static struct mlx5_ib_flow_prio *get_flow_table(struct mlx5_ib_dev *dev, log_max_ft_size)); fn_type =MLX5_FLOW_NAMESPACE_EGRESS ; prio = &dev->flow_db->egress_prios[priority]; -if (!dev->is_rep && +if (!dev->is_rep && !esw_encap &&MLX5_CAP_FLOWTABLE_NIC_TX (dev->mdev, reformat)) flags |=MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT ; }
@@ -3887,24 +3891,28 @@ _get_flow_table(struct mlx5_ib_dev *dev,int max_table_size =0 ;u32 flags =0 ;int priority; +u8 esw_encap;if (mcast) priority =MLX5_IB_FLOW_MCAST_PRIO ;else priority =ib_prio_to_core_prio (fs_matcher->priority, false); + esw_encap =mlx5_eswitch_get_encap_mode (dev->mdev) != +DEVLINK_ESWITCH_ENCAP_MODE_NONE ;if (fs_matcher->ns_type ==MLX5_FLOW_NAMESPACE_BYPASS ) { max_table_size =BIT (MLX5_CAP_FLOWTABLE_NIC_RX (dev->mdev, log_max_ft_size)); -if (MLX5_CAP_FLOWTABLE_NIC_RX (dev->mdev, decap)) +if (MLX5_CAP_FLOWTABLE_NIC_RX (dev->mdev, decap) && !esw_encap) flags |=MLX5_FLOW_TABLE_TUNNEL_EN_DECAP ;if (MLX5_CAP_FLOWTABLE_NIC_RX (dev->mdev, - reformat_l3_tunnel_to_l2)) + reformat_l3_tunnel_to_l2) && + !esw_encap) flags |=MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT ; }else if (fs_matcher->ns_type ==MLX5_FLOW_NAMESPACE_EGRESS ) { max_table_size =BIT (MLX5_CAP_FLOWTABLE_NIC_TX (dev->mdev, log_max_ft_size)); -if (MLX5_CAP_FLOWTABLE_NIC_TX (dev->mdev, reformat)) +if (MLX5_CAP_FLOWTABLE_NIC_TX (dev->mdev, reformat) && !esw_encap) flags |=MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT ; }else if (fs_matcher->ns_type ==MLX5_FLOW_NAMESPACE_FDB ) { max_table_size = BIT(
Signature · 1 line
2.20.1
Re: [PATCH rdma-next 2/3] RDMA/mlx5: Consider eswitch encap mode
Hi,
Quoted 9 lines
From: Maor Gottlieb <[email protected]>
When flow steering is created, then the encap support should consider the eswitch encap mode. If the eswitch flow table (FDB) supports encap then it shouldn't be supported on NIC RX flow tables.
Fixes 4adda1122c490 ('RDMA/mlx5: Enable decap and packet reformat on flow tables')Signed-off-by Maor Gottlieb <[email protected]>Signed-off-by Leon Romanovsky <[email protected]>
Kind regards,
Petr
PATCH 3/3 RDMA/mlx5: Enable decap and packet reformat on FDB
In reply to Leon RomanovskyFrom: Maor Gottlieb <[email protected]>
If FDB flow tables support decap operation, enable it on creation, This allows to perform decapsulation of tunnelled packets by steering rules. If FDB flow tables support reformat operation, enable it on creation as well.
--- drivers/infiniband/hw/mlx5/main.c | 5 +++++ 1 file changed, 5 insertions(+)
drivers/infiniband/hw/mlx5/main.c +5 -0
@@ -3917,6 +3917,11 @@ _get_flow_table(struct mlx5_ib_dev *dev, }else if (fs_matcher->ns_type ==MLX5_FLOW_NAMESPACE_FDB ) { max_table_size =BIT (MLX5_CAP_ESW_FLOWTABLE_FDB (dev->mdev, log_max_ft_size)); +if (MLX5_CAP_ESW_FLOWTABLE_FDB (dev->mdev, decap) && esw_encap) + flags |=MLX5_FLOW_TABLE_TUNNEL_EN_DECAP ; +if (MLX5_CAP_ESW_FLOWTABLE_FDB (dev->mdev, reformat_l3_tunnel_to_l2) && + esw_encap) + flags |=MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT ; priority =FDB_BYPASS_PATH ; }
Signature · 1 line
2.20.1
Re: [PATCH rdma-next 3/3] RDMA/mlx5: Enable decap and packet reformat on FDB
Hi,
Quoted 9 lines
From: Maor Gottlieb <[email protected]>
If FDB flow tables support decap operation, enable it on creation, This allows to perform decapsulation of tunnelled packets by steering rules. If FDB flow tables support reformat operation, enable it on creation as well.
Signed-off-by Maor Gottlieb <[email protected]>Signed-off-by Leon Romanovsky <[email protected]>
Kind regards,
Petr